Meta planned to cut 60% of some teams. Then AI showed it couldn't do the job.

Meta's "Project OT" was the most ambitious AI workforce replacement plan in corporate history.

60% of some teams. Two waves. May and November.

Then the data came back.

Code changes surged 220% year-over-year. But features reaching users? Only up 36%. The AI was generating noise, not output.

Security incidents spiked 40%. Time spent firefighting jumped 70%. Unchecked AI agents were performing "large-scale, disruptive actions that humans are unlikely to execute."

Employee sentiment cratered from 74% to 55%. Staffers flooded internal channels with elephant emojis — the layoffs were the elephant in the room.

Zuckerberg called off the second wave hours before it launched. May 20 still cut 10%. But November died.

The cruel irony: Meta's total employee compensation rose 30% in H1 206. The AI researchers replacing the old workforce cost more than the people they replaced. And Meta is burning $130B+ on AI infrastructure this year alone.

This isn't a failure of ambition. It's a failure of math. The company that bet the hardest on AI replacing humans just proved the tech isn't ready.

If Meta can't make it work with their resources, your AI transformation plan needs a serious reality check. Audit your AI agent deployments today. Measure actual output, not lines of code. The gap between "AI wrote it" and "AI shipped it" is where your budget goes to die.